Earlier we reported (Basden, Basden, Bryner, & Thomas, 1997) that, in comparison with nominal groups (three people tested individually), three-person collaborative groups recalled fewer presented words but intruded more nonpresented words. In the present research, Experiment 1 showed that when presented words were associatively related to critical nonpresented words, collaboration inhibited recall of presented words but did not influence recall of critical nonpresented words. Experiment 2 showed that with categorized lists, recall of high taxonomic frequency critical nonpresented words was greater for collaborative groups than for nominal groups. Collaboration did not inhibit recall of presented words, presumably because guessing supplemented recall in collaborative groups. Greater false recall in collaborative than in nominal groups appears to result from activation of superordinate-to-item associations rather than item-to-item associations.
MEMORY DISTORTION IN GROUP RECALLIn reviewing previous research, our concern was with the effect of the collaborative efforts of several individuals on the overall accuracy of their recall. Perhaps the earliest research reporting that group remembering introduces memory distortions came in the form of Bartlett's (1932) classic descriptions of reconstructions in recall. Bartlett suggested that memory changes occurred in conjunction with the development of schemas, socially influenced representations of the gist of the passage. He used the Method of Serial Reproduction, in which one individual first reads a prose passage such as the North American folktale "The War of the Ghosts," and then relates it from memory to a second individual, who in turn relates it to a third individual, and so on. Bartlett did not compare distortions by individuals repeatedly recalling alone (The Method of Repeated Reproduction) with those by individuals who related the passage to others in a group (The Method of Serial Reproduction). However, he did report dramatic changes in content with both methods. Thus, Bartlett's studies, as well as Allport and Postman's (1945) studies of rumor transmission in groups, indicate that collective recall may reduce accuracy.Perlmutter (1953) extended Bartlett's (1932) research by testing the recall of collaborating two-and three-person groups twenty-four hours after they had recalled "War of the Ghosts" as individuals. Perlmutter reported that individuals tended to reproduce the same errors they had introduced earlier, and that these errors cumulated across the group to make the product of collaborative recall less accurate than that of individual recall. Similar findings were reported by Perlmutter and de Montmollin (1952). Stephenson, Brandstatter, & Wagner (1983) compared initial trial recall of "War of the Ghosts" by individuals and two-person groups. They concluded that twoperson groups showed more reconstructive errors and were more confident of those errors than individuals. In contrast, Yuker (1955) reported that group recall was more accurate than indiv...